Monoclonal antibodies to plant nuclear matrix reveal intermediate filament- related components within the nucleus

The term ‘nuclear matrix’ has been widely used to denote those nuclear proteins that remain after extraction of nuclei with nonionic detergents, high molarity salt solutions and treatment with nucleases (either DNase or more specific nucleases alone, or in combination with ribonuclease: for reviews see, e.g., Nigg, 1989, 1988; Newport and Forbes, 1987; Gerace and Burke, 1988). Although the validity of a definition based on particular extraction procedures may be questioned, evidence is accumulating that this class of proteins has a major role in many aspects of nuclear activity (see, e.g., Van der Velden and Wanka, 1987). In particular, several groups of workers have identified specific DNA sequences that attach to the insoluble matrix components (reviewed by Mirkovitch et al. 1987). These sequences show homology to the consensus cleavage sequence for DNA topoisomerase II, suggesting a role for this protein in maintaining and modifying higher-order chromatin structure.

There is however considerable controversy about how real an internal nuclear matrix may be in structural terms. Thus although fibrous networks have been demonstrated inside the nucleus after extraction with nonionic detergents, salt and nucleases (see e.g. Fey et al. 1986), such studies have been criticised on the grounds that these treatments could cause artefactual precipitation and nonspecific polymerisation of the proteins present (Cook, 1988). The use of ionic conditions as near as possible to those in the cell in the extraction buffers goes some way to answering these objections. This approach has been adopted by Jackson and Cook (1985, 1988), who encapsulated cells in agarose microbeads, then treated them with nonionic detergents in an isotonic buffer. Finally, DNA was removed by nuclease digestion and electrophoresis. Electron microscopy then showed the presence of an extensive filamentous network. A repeat spacing of 23 nm was visible along the fibres, strongly suggesting a relation to cytoplasmic intermediate filaments (IFs).

An alternative approach is to analyse the matrix proteins by biochemical and immunological techniques and then to determine their localisation subsequently by immunofluorescence and immunogold microscopy. In this way several nuclear matrix proteins have been identified, most notably the nuclear lamins (Krohne and Benavente, 1986). This group of proteins has been analysed in considerable detail. Sequence analysis (McKeon et al. 1986) as well as antigenic crossreactivity (e.g. see Lebel and Raymond, 1987; Osborn and Weber, 1987) has shown that they are closely related to the cytoplasmic intermedi- ate filament proteins, sharing with them the overall molecular organisation of a central alpha-helical domain, with nonhelical head and tail regions. Importantly, lamins also contain the consensus amino acid sequence detected in almost all IFs. Lamins have been identified in several vertebrate and invertebrate animal species and recently in yeast (Georgatos et al. 1989). They form a structure called the lamina - a thin fibrous network just within the inner nuclear membrane. Electron microscopy of an oocyte lamina (Aebi et al. 1986) has shown it to comprise filaments of similar dimensions to cytoplasmic IFs in a criss-cross arrangement. It has been suggested that the lamina is important in mediating interactions between chromatin and the nuclear envelope. The finding of different lamin isotypes during development in amphibians (see Krohne and Benavente, 1986) as well as in other vertebrates (Lehner et al. 1987; Stewart and Burke, 1987) has suggested a role for the lamins in determining celltype specific chromatin structure, which in turn might be a mechanism for differential gene expression (Nigg, 1988). The lamins are known to be phosphorylated (Ottaviano and Gerace, 1985); their hyperphosphorylation is thought to be part of the cascade of events leading to nuclear envelope breakdown during cell division.

Most investigations of the nuclear matrix relate to mammals, amphibians or insects (specifically Drosophila melanogaster). There is relatively little information for any other species and virtually none for the plant kingdom. There has been a report of a nuclear matrix prepared from Allium cepa L. (Ghosh and Dey, 1986). Galcheva-Gargova et al. (1988) have reported the isolation of a nuclear ‘shell’ from Zea mays L. and Phaseolus vulgaris L. In this preparation they found a rather simple profile of proteins of approximate M_r 45–65(×10³) and showed by immunoblotting analysis a possible relation between these proteins and animal IF/lamina proteins. Recently, Worman et al. (1989) showed cross-reactivity of an antibody to turkey lamin B with pea nuclei, labelling a band of 65×10³M_r on immunoblots. Corben et al. (1989) have identified a nucleolar matrix protein by monoclonal antibody techniques. Sanchez-Pina et al. (1989) have used a variety of antibodies directed against animal small nuclear RNP proteins and nuclear matrix proteins, and have shown by immunocytochemical techniques crossreactivity in carrot suspension culture cells. Several studies have now confirmed that plants contain cytoplasmic IF antigens (Dawson et al. 1985). These are recognised by IFA. Plant cells also contain cytoplasmic fibrillar bundles that share many similarities with IFs, including reconstitution from urea (Hargreaves et al. 1989b). An antibody (AFB) against these cytoplasmic bundles recognises type IH animal IFs (Hargreaves et al. (1989a). However, the evidence for nuclear IF-related proteins in plants is still not on solid ground.

We describe here the preparation of a nuclear matrix fraction from carrot (Daucus carota L.) suspension culture cells, its use to raise a library of monoclonal antibodies, and the initial characterization of two of the antibodies. Further, we show that the nuclear proteins recognised by these antibodies are arranged as fibres within and at the periphery of the nucleus, and show immunological relatedness to lamin/IF proteins.

Unless otherwise stated, all chemicals were obtained from Sigma Chemical Co. Ltd, Fancy Road, Poole, Dorset, UK. The non- embryogenic carrot cell line described by Lloyd et al. (1980) (Daucus carota L, cultivar Scarlet Nantes) was used. Cells were grown in suspension culture in Murashige and Skoog’s medium (GIBCO) containing 3% (w/v) sucrose, 0.5 mg l⁻¹ kinetin and l.0 mgl⁻¹ 2,4-dichlorophenoxyacetic acid (2,4-D), at 25°C with a 12 h light/dark cycle with agitation at 120 revs min⁻¹ on an orbital incubator. Cells were used 8–10 days after subculturing. The cells were pelleted from the growth medium at 100 g and then resuspended in fresh medium to which had been added 0.5 M mannitol and 0.4% (w/v) Driselase. The cells were then incubated overnight in the previous growth conditions to produce protoplasts.

All preparative procedures following cell breakage were carried out on ice, and the following protease inhibitors were included in all buffer solutions: 10 μM benzamidine hydrochloride, 1 mM phenyhnethylsulphonyl fluoride (PMSF), 1/zgml^-1 phenanthroline, 10 μg ml⁻¹ aprotinin, 10 μg ml⁻¹ leupeptin, 10 μg ml⁻¹ pep- statin A. The protoplasts were gently pelleted at 100 g and resuspended in buffer Cl (0.5 M sucrose, 10 mM NaCl, 10 mM Mes (pH 5.3), 0.15 mM spermine, 0.5 mM spermidine). Intact, healthy protoplasts were purified by flotation in this medium after centrifugation at 150 g. The purified protoplasts were resuspended in Buffer C (as buffer Cl except with 0.4 M sucrose) to which was added 0.05% (v/v) Triton X-100. They were gently broken with 1-3 strokes in a stainless steel homogeniser to release the nuclei. After gently spinning down onto a Maxidens cushion (Nyegaard (UK) Ltd, Myken House, 11 Wagon lane, Sheldon, Birmingham) at 400 g, the nuclei were carefully resuspended in a small volume of the same buffer, layered onto buffer C containing 15% (v/v) Percoll, and spun (10 min, 500g) onto a Maxidens cushion. All stages were monitored by light microscopy, to ensure that pure, uncontaminated nuclei were obtained.

The nuclei were next suspended in buffer C+l % Triton X-100 and stirred on ice for 15 min, then spun onto a cushion of Maxidens. The nuclei were resuspended in 50 mM Mes (pH 5.3), ImM magnesium chloride, 50 μg ml⁻¹ DNase I (Worthington, Lome laboratories Ltd, P.O.Box 6, Twyford, Berkshire, UK), and incubated on ice for 30 min. Then the solution was made up to 1 M NaCl and stirred for a further 30 min. The nuclei were pelleted by spinning at 500 g for 10 min, then the DNase and salt treatment step was repeated, to give the nuclear matrix pellet. The nuclear matrix fraction was suspended in either 4M, 2M or 1 M urea in 10mM Tris-HCl (pH8.0), ImM EDTA and stirred for 30min on ice, then centrifuged at 10000 g to give urea-soluble nuclear matrix fractions.

Nuclei were prepared from rat liver by the procedure of Blobel and Potter (1966), and the pore complex-lamina fraction was isolated as described by Dwyer and Blobel (1976).

One-dimensional (1-D) gel electrophoresis using 10% gels was carried out according to the procedure of Laemmli (1970). Samples were dissolved in the standard sample buffer and boiled for 1 min. Two-dimensional (2-D) electrophoresis was carried out as described by O’Farrell (1975). Electrophoresed proteins were transferred to nitrocellulose paper using the method described by Towbin et al. (1979). The nitrocellulose blots were first incubated for at least lh in blocking solution, 3% (w/v) bovine serum albumin, 2% (w/v) dried milk powder (Marvel), 2% foetal calf serum (v/v) in TBS (10 mM Tris-HCl, 140 mM NaCl, pH 7.4). Then the blots were incubated for either 1 h at room temperature or overnight at 4 °C in primary antibody, washed thoroughly with TBS, blocked again for 15 min and then incubated in secondary antibody coupled to horseradish peroxidase diluted in blocking solution. Antibody-labelled bands were visualised by the peroxidase reaction with 4-chloronaphthol (Hawkes et al. 1982). Total protein on gels was stained with Coomassie Brilliant Blue or with silver staining (Oakley et al. 1980), or on nitrocellulose blots by biotinylation and reaction with streptavidin-peroxidase (Biorad Laboratories Ltd, 32nd and Grittin Ave., Richmond, CA, USA).

A library of monoclonal antibodies was produced against the nuclear matrix fraction. The nuclear matrix pellet was dissolved in 1% SDS, insoluble material was removed by centrifugation, and the proteins were precipitated with 90% acetone, to produce an SDS-denatured antigen. This material was then used for immunisation of Lou-C rats. Immunisation protocols and hybrid- oma production and cloning were carried out according to standard procedures (Galfre et al. 1979; Galfre and Milstein, 1981). For primary screening of the hybridoma supernatants immunoblotting against whole nuclear extracts was used. Positive hybridomas were cloned by limiting dilution and characterised further by immunoblotting and immunofluorescence. The immunoglobulin class of all the monoclonal antibodies was determined by the precipitin reaction using a kit of standards (Serotec, 22 Bankside, Kidlington, Oxford 0×5 1JE, UK).

Other monoclonal antibodies used were: IFA, which is the monoclonal antibody described by Pruss et al. (1981), and has been shown to recognise all intermediate filament protein classes by interacting with the C-terminal region of the conserved domain (Geisler et al. 1983). AFB: anti-fibrillar bundle antibody. This monoclonal antibody was produced by Hargreaves et al. (1989a). It was raised against the fibrillar bundles of insoluble protein from carrot suspension culture cells, and recognises a plant cytoplasmic intermediate filament-related antigen, as well as vimentin from a range of animal cells and representatives of type HI animal intermediate filaments. ME101: anti-intermediate filament monoclonal antibody raised against peripherin (Escurat et al. 1989). It recognises all the major classes of vertebrate intermediate filament proteins in immunoblotting assays, although its reaction with lamins has not been characterised so far. It recognises specifically the amino-terminal end of the rod domain in intermediate filaments. LSI: anti-lamin A/C; human autoimmune antiserum (McKeon et al. 1983).

Nuclei were fixed in 1.3% (w/v) formaldehyde (freshly made from paraformaldehyde) in 10 mM Mes, 20 mu NaCl, 2mM EDTA, 130mM KC1, 0.05% (v/v) Triton X-100 (pH5.3) or in lOnm phosphate, 20 mM NaCl, 2mM EDTA, 130 mM KC1, ImM dithiothreitol (DTT), 0.05% (v/v) Triton X-100 (pH 5.3), which were judged to be isotonic by observation of the nuclei by phasecontrast microscopy during fixation. Alternatively, protoplasts were directly broken into one of the above buffers to minimise the possibility of damage during preparation. After fixation they were washed in buffer minus fixative and allowed to settle onto microscope slides that had been pretreated with poly-L-lysine and allowed to air-dry. The adhering nuclei were then gently washed with TBS. In some cases the slides were dipped for a few seconds into methanol at −20°C at this stage. Primary antibody was then added. The slides were incubated with the primary antibody either for lh at room temperature or overnight at 4°C. In some experiments the air-drying step was omitted. After washing and incubation with fluorescein-labelled secondary antibody, the samples were counterstained for DNA with 4,6-diamidino-2- phenylindole (DAPI) and mounted in antifade mountant (Citif- luor Ltd, City University, London EC1V OHB, UK). Microscopy was carried out on a Zeiss universal microscope equipped with epifluorescence optics. Secondary antibodies were obtained from Dakopatts, P.O.Box 1359, DK2600, Glostrup, Denmark.

Rat liver nuclei were fixed for immunofluorescence in 3.7% (w/v) formaldehyde (freshly prepared from paraformaldehyde) in 50 mM Tris-HCl (pH 7.5), 25 mM KC1, 5 mM MgClj. After fixation they were washed in buffer without fixative, allowed to settle onto poly-L-lysine-treated slides and then to air-dry. Antibody incubations were carried out as described as above.

Pea root vibratome sections (40 /on) were cut using a Micro-cut H1200 (Bio-rad). Sections were extracted sequentially with 1% Triton X-100 in MTSB (50mM Pipes, 5 MM EGTA, 5 HIM MgS0₄, pH6.9), 100 μg ml⁻¹ DNase in MTSB and 1 M NaCl in MTSB. In some experiments, the isotonic buffer described above was used instead of MTSB: no difference in labelling was observed. In some cases sections were immersed at room-temperature in MeOH, but this did not make any obvious difference. Antibody incubations were as decribed above.

Specimens labelled as above for immunofluorescence were analysed by confocal laser scanning microscopy. For this a BIORAD MRC500 confocal laser scanning head attached to a Zeiss universal microscope was used. The microscope fine- focusing was controlled by the computer, allowing focal section stacks to be collected. For these studies a Leitz Planapo 63 × objective was used (oil immersion, n.a. 1.4). The confocal pinhole was closed as far as possible, but generally the signal was too weak with the pinhole fully closed and the data sets were collected with a midway position. Focal sections were spaced at 1 μm or 2 μm intervals. For further processing and analysis the data sets were transferred to a Stardent Titan graphics workstation via an ethemet link. This computer was used for interactive display of section stacks (Agard et al. 1989). Images were photographed directly from the display screen.

We used a method for nuclei isolation similar in general terms to that reported by Saxena et al. (1985); for example, we found the use of a low pH (5.3) to be necessary. Several other factors were also of crucial importance. First, the protoplast preparation had to be of excellent quality; the optimal time for harvesting the cells was 8 days after subculturing. Cultures significantly older than this gave much lower yields, mainly because a higher proportion of the protoplasts did not float up into buffer Cl. In order to prevent the nuclei aggregating, a low concentration (0.05%, v/v) of Triton X-100 had to be added to the nuclear isolation buffer (buffer C). It was also important to exclude magnesium ions from this buffer. Initially we added the polyamines spermine and spermidine (see Mitchison and Sedat, 1983) to stabilise the chromatin in the absence of magnesium, but later experiments showed that this was not necessary, although it may increase the yield slightly. It was of great importance to handle the isolated nuclei extremely gently. Thus all centrifugation was at the lowest practicable speed for the shortest possible time, and the nuclei were always spun onto a cushion of Maxidens. Even with these precautions a deterioration in quality as judged by phase-contrast microscopy was apparent after more than two cycles of centrifugation and resuspension. It was also extremely important to include the entire cocktail of protease inhibitors (see Materials and methods). Without these, or with only PMSF and/or leupeptin, virtually all the protein bands at high molecular weight (greater than ∼50×10³) were lost. Thus these plant nuclei are very easily damaged both mechanically and by protease action.

Once purified nuclei were obtained, a fairly standard nuclear matrix preparation was followed. The nuclear membranes and residual endoplasmic reticulum (ER) were removed by washing with 1% (v/v) Triton X-100. Then DNA was digested by treatment with DNase I and the histones removed by washing with 1 M NaCl. Finally, in addition to examining the residual nuclear matrix pellet directly by SDS-PAGE, we also used differential extraction with 1, 2 and 4 M urea. A typical nuclear preparation viewed with phase contrast is shown in Fig. 1. The nuclei are virtually all intact and little or no other material is present. The nuclear matrix preparation (Fig. 2) retains most of the overall nuclear morphology. The extracted nuclei clearly have a good deal of internal structure, and the nucleoli are still intact and in place, although virtually all the DNA has been removed as judged by fluorescence microscopy after staining with DAPI. (The nucleolar structure is also maintained after treatment with RNase, which, however, we have not used routinely.)

SDS–PAGE profiles of the various fractions are shown in Fig. 3. The nuclear matrix pellet fraction (lane b) is clearly a complex mixture of many polypeptides. None of the components is present in a very much greater relative abundance than the others. This is in marked contrast to similar preparations from many vertebrate sources, which contain a large proportion of the lamin proteins. However, urea extracts, notably 2 M urea (lane d), show enrichment in several bands, in particular those at 53, 60, 92 and 120×10³Af_r and a group of bands at higher relative molecular mass. (An additional band at about 35×10³M_r arises from DNase carried over from the previous extractions.) 1 M urea preferentially solubilises the protein at 60×10³Af_r (lane e).

The nuclear matrix fraction was used to inoculate rats for monoclonal antibody production. Immunoblotting of the culture supernatants against SDS-PAGE blots of total nuclear proteins was used as the primary screening test. Cultures which show’ed positive blotting against only one or a few bands were selected for cloning. Of these, we report here the first characterisation of two (JIM 62 and JIM 63). Both were of isotype IgM, as determined by precipitin reaction against a set of standards.

On 1-D immunoblots of whole nuclear extracts JIM 62 and JIM 63 labelled a single high molecular weight band (sometimes resolved into a doublet). Immunoblot labelling of the 4 M urea extract with JIM 62 and JIM 63 showed that that both recognise the 92×10³M_r protein(s) enriched in this extract (Fig. 4). These antibodies were characterised in greater detail on 2-D immunoblots (Fig. 5). Fig. 5 shows total protein staining on 2-D PAGE blots of whole cells (A), whole nuclei (B) and nuclear matrix (C). Fig. 5D and E shows the labelling of whole nuclei blots with JTM 63 and JIM 62 (the equivalent total protein staining is Fig. 5B). Both antibodies specifically label the 92×10³M_r group of proteins (approx pl 5.5–5.9). It appears that these are two species of close molecular weight, both of which exhibit a number of charge isomers. On blots of whole cells, JIM 62 labels only the 92×10³M_r proteins. JIM 63 labels the 92×10³M_r proteins and also six other lower molecular weight spots. The staining of 2-D nuclear matrix and urea supernatant extracts with these antibodies was essentially the same as whole nuclei. Thus the 92xlO^ilAf_r proteins are nuclear components that are selectively enriched in the nuclear matrix fraction and thereafter in the urea supernatant. Since we were initially interested in searching for lamin and intermediate filament-related proteins we screened the immunoblots with antibodies to plant intermediate filament proteins (AFB, Hargreaves et al. 1989a), to anima] intermediate filament proteins (IFA, Pruss et al. 1981; and ME101, Escurat et al. 1989) and to human lamins A+C (LSI, McKeon et al. 1983). All these antibodies except IFA labelled the 92×10³M_r proteins. IFA labels the 60×10³M_r protein enriched in the urea supernatants. This is shown by 2-D immunoblotting (Fig. 5F, G, H).

Initial attempts at immunofluorescence staining using whole cells or protoplasts showed either weak diffuse staining throughout the cell (JIM 62) or staining of a network around the cell cortex and nuclear periphery (JIM 63), very similar to that reported for AFB (Hargreaves et al. (1989a). However, we noticed that where the cells had broken open to expose the nucleus or where there were isolated nuclei free of cytoplasm we obtained bright, specific intranuclear staining. Therefore, in order to eliminate the influence of the surrounding cytoplasm, we used isolated nuclei to probe the distribution of the JIM 62 and JIM 63 intranuclear antigens. The nuclei were prepared either as for matrix isolation, followed by fixation in formaldehyde, or by disrupting protoplasts directly into formaldehyde fixative in a modified ‘isotonic’ buffer (Jackson and Cook, 1985, 1988). Initial trials using the buffer described by Jackson and Cook for animal cells showed that it was substantially hypotonic for the carrot protoplast nuclei. We therefore kept the same ionic concentrations as this buffer and adjusted the formaldehyde concentration until it was osmotically balanced for these nuclei (1.3% (w/v) formaldehyde). In this way the nuclei were fixed with minimum disruption. The fixation was in all cases monitored by optical microscopy. The nuclei were then attached to po]y-L-lysine-coated slides and treated as for standard immunofluorescence. Equivalent results were obtained whichever method was used.

Examples of immunofluorescence labelling of carrot nuclei by JIM 63 are shown in Fig. 6. JIM 63 sometimes shows labelling of the nuclear rim; however, the most striking feature is labelling of intranuclear fibres or fibre bundles. They are sometimes reminiscent of cytoplasmic fibrillar bundles, but also comprise a fine internal network. In control experiments the primary antibody was omitted, or unrelated antibodies (e.g. YOL-34 antitubulin, Kilmartin et al. 1982) were used. These did not show any labelling. In another set of control experiments the nuclei were incubated with rhodamine-phalloidin in order to test the possibility that the labelling represented F actin. Again no labelling significantly above background was seen. In a further set of experiments the nuclei were treated as for nuclear matrix isolation, and the residual nuclear matrices were labelled for immunofluorescence. The labelling was as strong as or stronger than that for intact nuclei. This shows that the labelling observed is due to the matrix material in the nuclei. The structural preservation was much poorer, however, and so we have used intact nuclei for detailed immunofluorescence analysis.

Similar intranuclear fibres are labelled by the other five antibodies - JIM 62, ME101, AFB, IFA and LSI - however, the immunofluorescence labelling with JIM 62 and IFA is fairly weak. Examples of the other three antibodies are shown in Fig. 7. All nuclei show the fine network staining; the size and number of the fibre bundles are variable, although most nuclei show some trace of them. In some nuclei there are one or two large fibres extending right across the nucleus. In others there may be several short, stubby fibres. Often several fibres link together forming quite a complex network (see Fig. 6B and Fig. 7B). They are never seen to penetrate the nucleolus, although they sometimes appear to form a cage-like structure around it (e.g. Fig. 6A,B, Fig. 7C). They are always completely contained within the nucleus, and often extend to the nuclear rim. Goodbody et al. (1989) found that cytoplasmic staining with IFA and AFB differed according to whether the specimens were treated with methanol after formaldehyde fixation, presumably because of differential exposure of the epitopes recognised by these antibodies. Indeed, fibrillar bundles were never labelled with IFA and were labelled with AFB only after treating specimens with methanol. We therefore carried out a series of experiments where the nuclei were treated with methanol after paraformaldehyde fixation. Differences were only observed with JIM 63, which gave a much stronger labelling of the nuclear rim following methanol treatment (see Fig. 6C), in addition to the internal network. The intranuclear fibres were still seen with the other antibodies after methanol treatment.

JIM 63 shows wide cross-reactivity against other species. We have obtained very similar labelling of intranuclear networks and bundles with tobacco suspension culture nuclei, and nuclear rim and internal network labelling, but not bundles, with onion root nuclei. Immunoblot analysis has also shown labelling of bands at approximately 92×10³M_r in Arabidopsis thaliana, and preliminary results show cross-reactivity with a nuclear protein in Aspergillus nidulans (John Doonan and Graham Scofield, personal communication). We have also investigated the immunofluorescence labelling of JIM 63 on whole vibratome sections of pea root tissue, extracted as for nuclear matrix isolation. A series of confocal microscope sections of such a specimen is shown in Fig. 9. The most obvious feature is clear nuclear rim staining. In some cells the labelling extends into the nucleus, and there is also some cytoplasmic staining.

In order to test further the cross-reactivity of these antibodies, immunofluorescence labelling was carried out on rat liver nuclei with JIM 62 and JIM 63. For comparison, IFA and LSI were also used. Examples are shown in Fig. 8. As expected, IFA and LSI both show staining of the lamina (Fig. 8A,B). JIM 63 also shows a clear, bright staining of the nuclear rim. There is also a faint interior labelling (Fig. 8C). JIM 62 shows only weak interior labelling (Fig. 8D).

The plant nuclear matrix fraction we have prepared consists of a complex mixture of many polypeptides. In contrast, Ghosh and Dey (1986) obtained a matrix fraction displaying only three polypeptides between 55×10³M_r and 63×10³M_r. Galcheva-Gargova et al. (1988) also found only a very few proteins in this general range in their nuclear shell preparations. The difference may lie in the plant material used or in the preparative procedure. We found that omitting the cocktail of protease inhibitors, or using only PMSF or leupeptin, gave rise to severe proteolysis to the extent that all proteins above, and most of the minor bands below, 60×10³M_r, were lost.

We have identified a group of proteins of approximately 92×10³M_r in the nuclear matrix preparations by producing monoclonal antibodies to them, and have shown that these proteins can be selectively enriched by differential urea extraction. Furthermore, we have shown immunological relatedness between these proteins, intermediate filaments and mammalian lamins A and C using well- characterised monoclonal antibodies and an autoimmune serum of proven specificity. On 2-D blots, immunolabelling with both our monoclonal antibodies and the intermediate filament antibodies showed the 92×10³M_r proteins to comprise two rows of spots at closely similar molecular weights. This may be due to the presence of a series of charge isomers of two closely related polypeptides. An obvious possibility for such charge isomers would be protein phosphorylation. IFA labelled a more acidic polypeptide at approximately 60×10³M_r on 2-D immunoblots. It is possible that this protein is the analogue of the lamin B-type protein identified in pea roots by Worman et al. (1989).

These antibodies showed immunofluorescence staining of an intranuclear network in carrot suspension cell nuclei; this network often included the nuclear and nucleolar periphery, and in many cases bright, thicker fibres or fibre bundles were also observed. JIM 63 showed the clearest immunofluorescence and was studied in more detail. The nuclear rim labelling with JIM 63 is much stronger after cold methanol treatment. The fact that both JIM 63 and other antibodies still label fibre bundles after methanol treatment argues against the effect with JIM 63 being due to a difference in the location of the antigens when using the different fixation protocol. It seems more likely that this treatment alters exposure of the epitopes on the nuclear rim. Similar ‘unmasking’ of an epitope has been observed for several anti-intermediate filament antibodies previously. For example, Franke et al. (1983) showed that treatment with methanol, detergent or urea was necessary to stein PtK² cells with an anti-cytokeratin, and Pruss et al. (1981) found that acid-ethanol fixation was necessary for IFA. In thick vibrateme sections of pea roots extracted as for nuclear matrix isolation - i.e. in situ nuclear matrices - we also observed strong nuclear rim staining, in some cases accompanied by internal nuclear staining.

The fibre bundles we have observed within the nuclei are strongly reminiscent of the cytoplasmic fibrillar bundles found within the same cells (Powell et al. 1982), and it might be suggested, therefore, that we are seeing cytoplasmic bundles that have in some way become incorporated into the nucleus, perhaps during nuclear envelope re-formation after cell division. This is unlikely for several reasons. First, the proteins that comprise the cytoplasmic fibrillar bundles are very different in molecular weight from those we have identified, with no components as high as 92×10³M_r. Hargreaves et al. (1989b) showed by reconstitution studies that the major fibre-forming polypeptides occur at 58×1.0³M_r and 62×10³M_r. Furthermore, JIM 62 and JIM 63 do not give specific labelling of the major bands in immunoblots of purified cytoplasmic fibrillar bundles. Second, we often observe the nuclear fibre bundles in a rather complex network, which may enmesh the nucleolus completely. This suggests that formation of the fibre bundles may take place at the same time as or after formation of the nucleolus, i.e. after re-formation of the nuclear envelope. If this is the case we would expect to find a nuclear- targetting signal sequence in these proteins. Finally, Goodbody et al. (1989) found that AFB stained cytoplasmic fibrillar bundles only after methanol treatment, whereas the intranuclear fibres we have observed are labelled by AFB without methanol treatment. Several types of nuclear inclusions have been identified by electron microscopy in plant cells, both in cultured cells and in intact plants (for a review, see Wergin et al. 1970). Amongst these are inclusions that bear a striking resemblance to cytoplasmic fibrillar bundles on purely morphological criteria. It is possible that these inclusions represent at least the larger fibre bundles that we observe. The function of the nuclear inclusions is unknown (as is the function of the cytoplasmic fibrillar bundles) but it is possible that they are produced as a response to heat shock or other stress.

In summary, the 92×10³M_r proteins recognised by JIM 63 are immunologically related to intermediate filaments and show a distribution both within plant nuclei and at the nuclear periphery. This is different from the generally accepted view of vertebrate lamins. However, recent results show that, even in vertebrate nuclei, the lamins form a discontinuous network (Paddy et al. 1990). Thus it is possible that the 92×10³M_r proteins are analogous to vertebrate lamins A and C; alternatively, they may represent a new class of intermediate filament proteins. We have also shown that JIM 63 labels the nuclear rim in rat liver nuclei, suggesting that the antibody recognises one or more rat lamins. Preliminary evidence shows a cross-reaction of JIM 63 with rat liver lamins (together with some other bands) on immunoblots of rat liver nuclear matrices. We have also obtained preliminary results showing immunofluorescence staining of the nuclear rim and a vimentin-like cytoplasmic network in whole PtK₂ cells. Further studies are under way to try to define clearly the proteins in vertebrate cells that JIM 63 recognises. It will also clearly be necessary to extend our structural studies to the electron-microscope level with immunogold labelling. Amino acid sequence information for the proteins that we have identified is needed in order to characterise their relation to intermediate filament and lamin proteins, and to determine whether nuclear targetting sequences are present. Work is under way to achieve this.

We are grateful to Marc Kirschner for providing a sample of LSI, to Michel Escurat for ME 101, and to Clive Lloyd and Alan Hargreaves for providing AFB and for useful discussion. We thank Clive Lloyd, John Doonan and Keith Roberts for critical reading of the manuscript. We thank Peter Scott, Andrew Davies and Nigel Hannant for photography. This work was supported by the Agricultural and Food Research Council of the UK, via a grant-in-aid to the John Innes Institute.